Dissertation: The Commitment Credibility of Public-Employee Pensions
My dissertation concerns defined-benefit pensions for public employees. A lack of national regulation has facilitated localized influence in plan policies, creating a real potential for underfunding and unsustainable growth. Maintaining pension promises while keeping governments functioning effectively is one of the fundamental policy challenges of the 21st Century. How does split authority between elected officials and quasi-independent management-boards contribute to variation in plan governance and the ultimate provision of public goods and services?
Though elected officials have the final say over pensions, boards of trustees also influence plan governance. Not a great deal is known about boards or how they shape policies. Boards are composed of politically and non-politically appointed members, as well as active and retired employees. Plan active-employee size turns out to be the best predictor of membership, suggesting that employee voice expands as plans cover more workers. Using both fixed effects and instrumental variables approaches, I show how boards shape plans' policies and funded levels. Active and retired members shape discount rates, while active membership is positively associated with funded ratios. Interestingly, gridlock is also associated with higher discount rates. However, I find that plans' actual investment returns are poor predictors of expected returns, irrespective of board composition. While boards offer a venue through which states can manage funds, they are not suited to solving pensions' governance challenges alone.
While the payment of pension benefits is mandatory, states do not have to make their Annual Required Contributions (ARCs) into funds. I ask what factors are associated making these payments. Politicians appear especially committed to making payments into plans covering police and fire employees. They also make smaller contributions when pensions' other two revenue streams increase: investment returns and employee contributions. Thus, politicians prefer not to devote money to pensions when possible. However, they do commit to pensions when it is politically expedient, as is the case when demonstrating support for police and fire employees.
In theory, deferred pension income should encourage public employees to remain in their jobs for long and predictable periods of time, stabilizing retirement rates. However, it is unknown if and how pensions actually do that. Aside from pension generosity, employee representation on plan management boards could influence retirement rates. Representation provides workers with a sense of voice and influence over plan policies, encouraging them to remain in their jobs longer. This article uses a new proxy variable for retirement rates collected from 93 state-employee plans between 2001 and 2014. The results indicate that smaller salaries and larger pension benefits are associated with fewer retirements. At the same time, greater employee representation on boards is linked to reduced retirement rates, suggesting that both pension generosity and administration play key roles in public personnel management.
We develop and assess an elite-information account of representation. Politicians face uncertainty about voter opinion, and use previous vote-margins to gauge future electoral outcomes. Losses in vote support elicit ideological moderation given new information about electorates. To test this account, we use rain around Election Day as a natural experiment in voting in U.S. House races from 1956 to 2008. We find each additional inch of rainfall exogenously dampens Democratic vote-margins and shifts incumbents rightward in subsequent Congresses. We find responsiveness mainly in competitive districts, and by Democrats rather than Republicans, suggesting a party asymmetry in representation. Overall, we highlight the importance of elite information uncertainty in the electoral connection, and show that idiosyncratic electoral effects can meaningfully impact legislative behavior.
Unlike elections involving candidates, the Supreme Court has declared that referendum spending limits would infringe upon free speech, as money cannot possibly corrupt propositions. Nonetheless, four states have imposed limits. We exploit this variation to examine how money influences referenda policies. Using machine learning models for text analysis and empirical analyses of referenda finance data, we explore how laws shape the kinds of policies that become ballot initiatives. Specifically, we find that business interests play a substantial role in influencing the language of propositions, and that that influence is especially pronounced in states lacking donation caps.